Nucleotide diversity of the ZmPox3 maize peroxidase gene: Relationships between a MITE insertion in exon 2 and variation in forage maize digestibility

Background Polymorphisms were investigated within the ZmPox3 maize peroxidase gene, possibly involved in lignin biosynthesis because of its colocalization with a cluster of QTL related to lignin content and cell wall digestibility. The purpose of this study was to identify, on the basis of 37 maize lines chosen for their varying degrees of cell wall digestibility and representative of temperate regions germplasm, ZmPox3 haplotypes or individual polymorphisms possibly associated with digestibility. Results Numerous haplotypes with high diversity were identified. Frequency of nucleotide changes was high with on average one SNP every 57 bp. Nucleotide diversity was not equally distributed among site categories: the estimated π was on average eight times higher for silent sites than for non-synonymous sites. Numerous sites were in linkage disequilibrium that decayed with increasing physical distance. A zmPox3 mutant allele, carrying an insertion of a transposable element in the second exon, was found in lines derived from the early flint inbred line, F7. This element possesses many structural features of miniature inverted-repeat transposable elements (MITE). The mutant allele encodes a truncated protein lacking important functional sites. An ANOVA performed with a subset of 31 maize lines indicated that the transposable element was significantly associated with cell wall digestibility. This association was confirmed using an additional set of 25 flint lines related to F7. Moreover, RT-PCR experiments revealed a decreased amount of corresponding mRNA in plants with the MITE insertion. Conclusion These results showed that ZmPox3 could possibly be involved in monolignol polymerisation, and that a deficiency in ZmPox3 peroxidase activity seemingly has a negative effect on cell wall digestibility. Also, genetic diversity analyses of ZmPox3 indicated that this peroxidase could be a relevant target for grass digestibility improvement using specific allele introgressions.


Background
Because of its high energy content and good ingestibility, silage maize is a major source of forage for dairy cattle. Its energy value depends mostly on the breakdown of cell walls by rumen micro-organisms. Lignin interferes with the digestion of cell wall polysaccharides by acting as a physical barrier to microbial enzymes. In forage crops, lignin content, lignin structure and cross-linking between cell wall components influence digestibility [1].
Lignins are formed through dehydrogenative polymerization of two or three monolignols. Grass lignins are primarily made of two major units; guaiacyl (G, monomethoxylated on the aromatic ring) and syringyl units (S, di-methoxylated), resulting of the polymerization of coniferyl and sinapyl alcohols, respectively. The last step in lignin biosynthesis transport of the monolignols to the cell wall thus involves the oxidation of monolignols within the plant cell wall matrix. The dehydrogenation to monolignols radicals is attributed to different classes of enzymes, such as peroxidases (oxidoreductases, EC 1.11.1.7), laccases (oxidoreductases, EC 1.10.3.2) and (poly)phenol oxidases [2,3]. However, the effects of modifying these enzymes on cell wall lignification and digestibility have not been investigated extensively in grasses. In tobacco, down-regulation of a peroxidase led to transformants with a reduced lignin content of up to 40-50 % [4]. Down-regulation of an anionic peroxidase in transgenic aspen was associated with lower lignin content and modified lignin composition [5].
In plants, peroxidases belong to a multigene family. In maize, 13 peroxidase genes are described in The Maize Genetics and Genomics Database (http:// www.maizegdb.org [6]). Three maize cDNAs coding for three different peroxidases (ZmPox1, ZmPox2 and ZmPox3) were isolated from a 9-day old tip-less root library [7]. Expression patterns of these peroxidases suggested that only ZmPox2 and ZmPox3 were involved in lignification. ZmPox3 was mapped by Biogemma SAS (A Murigneux, pers. com.) in the bins 6.05/07, and therefore co-localizes with a major QTL cluster for lignification and cell wall digestibility [8,9].
The breeding of forage crops for higher digestibility can involve specific genetic resources devised through genetic engineering of lignin biosynthesis [1]. In addition, the identification of natural allelic forms of candidate genes that correlate with higher digestibility, and their subsequent use in marker assisted selection (MAS) schemes, may become an interesting breeding tool. Gene sequence diversity has been primarily studied to understand the impact of selection during maize domestication and to identify the levels and patterns of genetic variation in a large sample of maize loci [10][11][12]. However, Thornsberry et al [13] and Palaisa et al [14] demonstrated that nucleotide diversity analysis of candidate genes for flowering date and endosperm colour, respectively, allowed for the identification of alleles responsible for variation of these quantitative agronomic traits. Recently, in Eucalyptus globulus, Poke et al [15] found polymorphisms which could alter enzyme function in cinnamoyl CoA reductase (CCR) and cinnamyl alcohol dehydrogenase 2 (CAD2).
Allelic variation of the ZmPox3 peroxidase was studied in a set of 37 inbred lines or ecotypes representative of germplasm currently used in breeding programs and displaying a large range of variation in cell wall digestibility. A zmPox3 mutant allele, carrying an insertion of a transposable element in the second exon, was found in five lines and one ecotype. This element possesses many features of the MITE (miniature inverted repeat transposable element) class of mobile element [16]. This MITE insertion was further investigated to elucidate the relationship between this peroxidase mutant allele and cell wall digestibility.

ZmPox3 gene diversity
The ZmPox3 gene, amplified for 37 diverse maize inbred lines or ecotypes (Table 1), is approximately 1.7 kb long, and the coding region contains two small introns of about 127 and 111 bp (Fig. 1A). The number of polymorphic sites and indel events for the complete ZmPox3 maize gene is reported in Table 2, along with two estimators of nucleotide diversity π [17] and θ [18]. Values are given for the complete sequence, as well as 3'UTR, introns, silent, synonymous and non-synonymous sites. Fourteen diverse haplotypes were found. There were 31 SNPs (i.e. an average of one SNP every 57 bp), and 17 indels which representing 20% of the whole sequence length. Of the 17 indels identified, 29% were single-bp, 23% were doublebp and 11% contained three-bp. Nucleotide diversity was not equally distributed among site categories (Table 2). Non-coding regions had a higher diversity, especially in the introns as compared to the coding region, where nonsynonymous sites showed very low diversity. The estimated π value was on average eight times higher for silent sites than for non-synonymous sites. Introns were particularly rich in SNPs and indels. Intronic regions had an average of one SNP every 14 bp while exons had only one SNP every 82 bp. The two ZmPox3 introns have 15 indels varying in size from 1 to 12 bp, while the 3'UTR region has 3 indels with size ranging from 2 to 5 bp.
For five inbred lines (F7, and its progenies F226, F227, F324 and F7012) and one ecotype (Québec28), the only indel found in the coding region was identified in the second exon of the ZmPox3 gene. This insertion corresponded to a transposable element sharing many features of the MITE class of mobile element [16]. This element was short (321 bp), had 15 bp imperfect inverted repeats and was flanked by a 5 bp (CTCAG) direct repeat generated upon insertion.

Amino acid changes
According to the amino acid change classifications of Li et al [19] and Grantham [20], of the five non-synonymous SNPs found in ZmPox3, two changes were conservative (Val/Leu and Phe/Leu) and three were moderately conservative (Ser/Thr, Ala/Ser and Ala/Thr). The N terminal signal peptide (residue 1 to 22), and the active site as well as the heme-ligand site were not affected by any of these five amino acid replacements [7]. However, out of the 9 putative N-glycosylation sites identified [7] two were affected by one moderately conservative and one conservative amino acid changes.

Description of the ZmPox3 MITE insertion
The MITE insertion in the second exon disrupts the gene 129 bp downstream of the 3' splice site of the first intron ( Fig. 1A) by introducing a stop codon 75 bp after the beginning of the insertion (Fig. 1A). The conceptual translation of the mutant allele results in a truncated protein of

Linkage disequilibrium and recombination
There were many occurrences of significant linkage disequilibrium (LD) among ZmPox3 polymorphic sites (20.6% of the tested LD are significant at P < 0.001) (Fig.  2). After correction for multiple tests using the Bonferroni procedure, 11.7 % of LD remained significant (P < 0.001). LD concerned sites mainly located in the first intron. The transposable element (S0465) was in strong LD with six polymorphic sites located in the first two exons and in intron 1 and 2 (Fig. 2). These polymorphisms in the vicinity of the insertion site may have been linked to the mutation and inherited as such. LD decayed with increasing physical distance according to Remington's model ( [21], adjustment F (1.350) = 457, P < 0.0001) (Fig. 3). Three recombination events were detected in ZmPox3 when assessed as the minimum number of recombination events using the algorithm proposed by Hudson and Kaplan [22].

Relationship between ZmPox3 alleles and cell wall digestibility
ANOVA performed for a subset of 31 maize lines, for which digestibility values are known ( Table 1), indicated that the insertion of the transposable element (MITE) in the second exon was significantly associated with cell wall digestibility (P = 0.032). This insertion characterized a group of five inbred lines related to F7 and having a high cell wall digestibility (F7, F7012, F324, F227 and F226).
Based on the assumption that the lack of ZmPox3 peroxidase activity might lead to reduced or modified lignification, comparison of cell wall digestibility was investigated in 25 lines related to F7 and did or did not carry the zmPox3 mutant allele ( Table 3). The cell wall digestibility was significantly higher in the 11 lines with the zmPox3 mutant allele. Lines having the MITE insertion were also more related to F7 than the lines having the normal allele, but this difference was not significant and the MITE effect remained significant when the percentage of F7 genome was added as a co-variable (Table 3). Moreover, it appeared that two genetic backgrounds were favorable to cell wall digestibility in this set of early flint lines, independently of MITE insertion ; F7 (the Lacaune ecotype) and the germplasm from Argentina into which was bred F64 and is now found in lines F564, F286, F324, F7064 and F7065. Lines having a high cell wall digestibility (value ≥ 4), but not carrying the mutant allele, were all related to the Argentina germplasm (F286 and F564). In addition, line F324 carrying both the MITE insertion and Argentina germplasm, had a very high cell wall digestibility.
Organisation of the ZmPox3 gene

Expression profiles of the ZmPox3 gene and its mutant allele
To determine whether the mutant allele was transcribed, the ZmPox3 expression in F7012 (mutant zmPox3 allele) and Lan496 (normal ZmPox3 allele) were compared. Forward primer U19S1 located at the 5' end of the gene was combined either with primer U19R1 situated upstream or with primer U19MITEAS situated downstream of the MITE insertion in F7012 (Fig. 1A). The use of internal (18S) and external (pAW109) controls confirmed that both RNA quantity and RT-PCR were homogenous for samples obtained from stems with surrounding leaf sheaths of the upper or basal part of the plant (Fig. 4A, lines 2 and 3). No difference in signal intensity was observed between F7012 and Lan496 for the upstream part of ZmPox3 (Fig. 4A, line 4). On the contrary, the bands spanning the MITE insertion were barely visible in F7012 (Fig. 4A, line 5). This decrease for the mutant line could be explained by rapid RNA degradation of the nontranslated part of the mutant allele, though we cannot exclude a decreased PCR efficiency due to possible secondary MITE structures. For experiments on leaf blades, Linkage disequilibrium (LD) matrix among ZmPox3 informative polymorphisms (SNPs and indels) for 37 maize sequences  To obtain a more detailed expression profile of ZmPox3, its expression was studied using stem segments composed of nodes and the adjacent inter-nodes from F2 line at silking stage. ZmPox3 expression was higher in lower-nodes and inter-nodes than in upper-nodes (Fig. 4B). Intense expression occurred in nodes close to the ear which developed on node 11. In the ear leaf, expression was very low 10 days after pollination, higher after 21 days and low again 30 days after pollination. ZmPox3 peroxidase expression appeared important in lignifiying tissue, but seemingly not in the earlier stage of lignification of adult plants. In comparison, O-methyl transferase (AldOMT) expression was stronger in older nodes and inter-nodes and extended much further to the younger nodes and inter-nodes at the top of the plant. In contrast, its expression was lower than that of ZmPox3 in ear leaves (Fig. 4B). These data confirmed that the two genes are involved in different steps of the lignification process in maize.

Discussion
ZmPox3 nucleotide diversity was comparable to that previously reported for other nuclear genes in maize, and even slightly higher. For instance Tenaillon et al [23] and Rafalski [24] found one SNP approximately every 48 and 130 bp in the 3'UTR and coding regions, respectively. In ZmPox3, we counted on average one SNP every 57 bp.
Moreover, ZmPox3 had a nucleotide diversity unequally distributed among site categories. There were 26 SNPs for all the silent sites and only five amino acids replacements for 357 residues (1.4%). Although the low frequency of non-synonymous polymorphisms suggested that at least some coding region of theses genes are constrained, classical tests of neutrality showed no evidence of selection (data not shown). However, previous molecular diversity studies in maize have shown that departure from neutrality is rare [23]. A signature of positive selection could be revealed by an allele with unusually long-range LD [25]. In ZmPox3 however, LD decayed very rapidly with increasing physical distance, within 200 bp on average. This result is consistent with recent studies showing that LD typically decays rapidly within individual maize loci [10,21,23].
A moderately conservative amino acid changes, located in the second exon, affected a putative N-glycosylation site. This SNP was characteristic of a cluster of four high digestibility lines (F564, EP1, Wis94-443 and Wis93-3520) and was associated with cell wall digestibility (ANOVA; P = 0.03). It would be interesting to determine whether this amino acid replacement affects enzymatic activity. For example, in rice, Larkin and Park [26] have reported that two SNPs which caused amino acid substitutions in the waxy gene affected the endosperm apparent amylose content. The MITE insertion was also shown to be associated with cell wall digestibility (ANOVA; p = 0.032). Furthermore, a survey of an additional set of 25 lines related to F7 revealed significant association between MITE and cell wall digestibility. This MITE element of 321 bp is 93% identical to the 316 bp insertion observed in the B-M033 allele of the b gene, a regulatory gene involved in anthocyanin synthesis in maize [27]. It is also very similar to a transposon in intron 4 (92% identity) of the maize β-Dglucosidase gene (accession U60560), and to a 316 bp insertion in the 5'UTR region of the maize waxy gene (89% identity) [28]. It has been suggested that MITE insertions in non-coding regions such as the 5'UTR or 3'UTR of transcribed maize genes, provide regulatory sequences involved in transcription initiation or mRNA stability and are thus important tools of evolution [16]. Thornsberry et al [13] found significant association between flowering time traits and the 485 bp insertion in the 5' non coding region of the Dwarf8 gene caused by a putative MITE element. To our knowledge, the insertion of a MITE element in the second exon of ZmPox3 has never been previously reported in the coding region of maize genes. This insertion very likely resulted in transcription of a truncated and non-functional ZmPox3. Thereby, the high digestibility of numerous maize lines carrying the zmPox3 mutant allele could be considered as the result of a deficiency in the activity of this peroxidase probably involved in the lignin biosynthesis.
Decay of linkage disequilibrium with distance between nucle-otide sites for ZmPox3 Figure 3 Decay of linkage disequilibrium with distance between nucleotide sites for ZmPox3. Curve shows non linear regression of r 2 on distance, by using model described in Remington et al [21]. The regression coefficient b1 is 0.00462  As reported by de Obeso et al [7], ZmPox3 mRNA was detected at low levels in lignifying tissues of young maize roots. Our RT-PCR experiments highlighted that the expression of the ZmPox3 gene was weaker in the first steps of stem lignification and in younger parts of the stem, than in older parts of the stem. These results, the well-known role of laccases and peroxidases at the final steps of lignin biosynthesis [7] and the identification of a ZmPox3 ortholog in the Zinnia elegans model system of lignifying tissues (S. Guillaumie, pers. com.) reinforce the possibility of an involvement of this peroxidase in maize monolignol polymerisation. The expression profile was clearly distinct from that of AldOMT indicating that lignification involves precise spatial and temporal regulation of the numerous genes involved in the different steps of this complex process.

Conclusions
Analysis of allelic diversity in relation with cell wall digestibility values likely validated ZmPox3 as a plausible candidate gene for silage maize digestibility improvement. This peroxidase could have a direct effect on plant cell wall digestibility, even more so given that this peroxidase is colocalized with a cell wall digestibility and lignification QTL. The comparison between ZmPox3 alleles that differ mostly by the presence or absence of the MITE element offers an opportunity to evaluate the impact of this kind of mutation event, as it was extensively done for brown-midrib alleles [29]. Thus, either the targeted introduction of the mutant allele in different genetic backgrounds or the down-regulation in transgenic lines could confirm the usefulness of ZmPox3 MITE disruption in breeding maize for higher digestibility. Moreover, because it seems to be a rare event, the search for the ZmPox3 MITE insertion in genetic resources could also help in determining the phylogeny of maize lines and introductions to Europe.

Plant material
Thirty-seven inbred lines were chosen in order to display a large variation in maize cell wall digestibility and to represent a diversified sample of genotypes used both in European breeding programs and representative of commonly used US germplasm (Table 1). Maize plants were grown in vermiculite supplemented with a nutrient solution under a 16/8 h light/dark regime for 10 days. Because an abnormal zmPox3 gene was first found in F7012, and then in different parents of F7012 which were all related to the INRA flint inbred F7, the existence of this mutant allele was then investigated in 15 other private and public lines related to F7 (Table 3). RT-PCR investigations of ZmPox3 expression were done i) on successive nodes and inter-nodes n° 5 to 14 in plants of line F2 harvested at silking date, ii) on ear leaves of line F2 harvested 10, 21 and 40 days after silking, under field conditions, and iii) in basal and upper plant parts (stem and sheaths), and in the largest expended leaves of lines Lan496 and F7012 harvested 50 days after sowing in a greenhouse.

DNA extraction, primer design and PCR amplification
Genomic DNA was isolated from young maize leaves using the plant DNAEASY miniprep kit (Qiagen).
A pair of primers flanking the MITE insertion was designed (Fig. 1A) as U19MITES (forward; 5'-GGCACT-GGAGGCTCAGGGTGTGTT-3') and U19MITEAS (reverse; 3'-AGGAGACAACGCCGGGGCAC-5'). If a maize line had the mutant allele, a 0.3 kb fragment corresponding to the MITE element was amplified. On plants having the normal allele, no amplification occurred (Table 4 and Fig.  1A). Combinations between pairs of primers U19S1 / U19AS1 and U19MITES / U19MITEAS allowed us to distinguish between homozygote and heterozygote genotypes ( Table 4). The procedure to detect the zmPox3 mutant allele has been also described in detail in the French patent FR0302954.
PCR amplification reactions were performed in 50 µl containing 100 mM tris-HCl, 500 mM KCl, 15 mM MgCl 2 and 0.01% gelatin, 200 µM of each dNTP, 0.2 µM of 5' oligo, 0.2 µM of 3' oligo and 2.5 units of REDTaq DNA Genomic Polymerase (Sigma). 100 ng of genomic DNA was used as template. To amplify the ZmPox3 gene, the program was 5 min at 95°C followed by 30 cycles of 30 s at 95°C, 30 s at 60°C, and 1.5 min at 72°C, followed by 5 min at 72°C. To genotype the MITE insertion in ZmPox3, the PCR amplification protocol was slightly modified and consisted of 5 min at 95°C followed by 25 cycles of 30 s at 95°C, 30 s at 65°C, and 30 s at 72°C, followed by 5 min at 72°C.

DNA sequencing
Sequencing was performed for each PCR fragments in both directions by Isoprim (France) and MWG-Biotech (Germany). Ecotype Québec28 was heterozygous at ZmPox3 locus and the two alleles have been sequenced. The sequences containing singletons were checked by reamplifying genomic DNA and partially re-sequencing the appropriate alleles. Sequences were aligned using CLUSTALW [30]. Sequence data from this article have been deposited with the EMBL/GenBank Data Libraries under accessions n° AY500781-AY500812; AY508159-AY508163 and AY508516.

RT-PCR analysis
For isolation of total RNA, tissues were placed together with a stainless steel bead in 2 ml Eppendorf tubes and frozen in liquid nitrogen. Using pre-cooled holders the tissues were ground to powder in a Mixer Mill MM300 (Qiagen) by shaking two times 30 sec. The ground, frozen tissues were vortexed with 1 ml TRIzol ® Reagent (Invitrogen) at room temperature until melted. The aqueous phase resulting from a 10 min centrifugation at 18000 g and 4°C was re-extracted with 200 µl chloroform at room temperature. The RNA was precipitated with 500 µl isopropanol for 10 min at room temperature. The RNA pellet obtained by 10 min centrifugation at 18000 g and 4°C was washed with 1 ml 70% ethanol, dried and resuspended in 30 µl RNAse free water. After treatment with RNase free DNase and inactivation of the DNAse according to the instructions of the supplier (AMBION) the RNA was quantified in a spectrophotometer at 260 nm. Approximately 5 µg of total RNA were reverse transcribed using random hexamers (Amersham) and reverse transcriptase without RNaseH activity (Fermentas). The 20 µl reverse transcription reaction also contained 2.5 × 10 5 copies of GeneAmplimer pAW109 RNA (Applied Biosystems). The obtained cDNA was diluted 50 times in water and 5 µl used for amplification by PCR in a volume of 20 µl.

Evaluation of maize silage digestibility
In vitro cell wall digestibility of lines (per se values) was investigated through different multi-year and multi-local experiments managed by the different partners. Values of cell wall digestibility were estimated as described in Roussel et al [8] through the DINAGZ criterion (in vitro digestibility of the "non starch (ST), non soluble carbohydrates (SC) and non crude protein (CP) part") based on the enzymatic solubility of the whole plant [31]. Data were gathered in order to obtain an index score of cell wall digestibility for each line from 1 (40 %) to 5 (60 %), according to the DINAGZ trait, whose heritability is most often close to 0.75 [1]. Lines F271 and F4, possessed the lowest and highest cell wall digestibility in this set of maize lines, respectively Data analysis DNA sequences were analyzed using DnaSP [32]. Levels of nucleotide diversity were estimated as mean pairwise differences (π ; [17]) and number of segregating sites (θ ; [18]). The minimum number of intragenic recombination events was estimated using the four-gamete test [22]. Insertions and deletions (indels) were excluded from the estimates. The significance of pairwise linkage disequilibrium (LD) among polymorphic sites (SNPs and indels) was tested using Fisher's exact test excluding non-informative sites (singletons), and corrected for multiple analyses using the Bonferroni procedure [33]. The decay of LD with physical distance along genes was evaluated by nonlinear regression (PROC NLIN in SAS software, SAS institute 1999) following Remington's model [21] that considers potential low mutation rate and adjustment for sample size.
Each informative polymorphic site (SNPs and indels) was examined independently for an association with variation in cell wall digestibility using an ANOVA. The set of 31 lines for which digestibility values are known was used (Table 1).
A single factor variance analysis (MITE presence or absence) was used to compare the average digestibility value of lines, either having or not having the zmPox3 mutant allele, in the set of 25 lines related to F7 (Table 3). MITE effect was similarly investigated after adding the percentage of F7 genome in each line (expected value according to the pedigree) as a co-variable.

Authors' contributions
CGC carried out most of the experimental studies and data analysis and participated in the discussion and preparation of the manuscript. CBT participated in the DNA sequencing and data analysis. DM carried out linkage disequilibrium measurement and participated in the discussion. PR carried out RT-PCR experiments. JR carried out the ZmPox3 cloning. AM carried the ZmPox3 mapping and participated in the design. JPM participated in the design and the coordination of the study. YB contributed to the design and coordination of the study and participated in the discussion and preparation of the manuscript.